Next Platform TV: October 9, 2020

On today’s show, in depth with Google’s lead for technical infrastructure on the topic of open infrastructure; a look at the Arm ecosystem in supercomputing and large-scale datacenters with computing pioneer, Simon McIntosh-Smith; we check in with Matt Kixmoeller of Pure Storage to discuss containers and flash–and all the sticky points in between; the show closes with an in-depth on IBM’s opening of its Power architecture and what it means for their competitive advantage. 

We start off with Urs Hölzle, senior vice president of the Technical Infrastructure team at Google and the necessity of open platforms in the datacenter and the need to “consumerize” IT so it is not so tough and costly to manage. We also spoke about what happens when companies do not differentiate so much on their ability to run IT infrastructure when it all gets easier and the playing field is level – and what makes Google think it knows how to make IT easier.

We also got an update from Simon McIntosh-Smith, professor of high performance computing, about the new Isambard 2 supercomputer being installed at the University of Bristol. This will be the second largest Arm-based supercomputer in the world, including a mix of processors and, importantly, will be a testbed to show how Arm can and does hold its own against anything the X86 market has when it comes to HPC codes.

We also spoke to Matt Kixmoeller, chief strategy officer at Pure Storage, about its acquisition of Portworx, why storage is such a pain in the neck for containerized environments, and what is going on out there in the datacenter when it comes to flash storage.

In the wake of the OpenPower Summit, we had a chart with Mendy Furmanek, president of the OpenPower Foundation and a member of IBM’s Power processor team, about the opening up of the Power instruction set and two core designs by IBM. The question now is can an opened Power architecture make headway against Arm and keep RISC-V at bay?

Thanks as always for watching. Next week we will be talking about MLPerf for HPC/AI; check in with the Texas Advanced Computing Center and its HPC director, Dan Stanzione; take a look at the evolving Spectrum Scale strategy for large-scale file system background; and much more. Tune in on Fridays or watch over the weekend.